Low Complexity and High Robustness Image Dehaze Algorithm

Abstract

This paper proposes a low complexity, high robustness dark channel prior image dehaze method. This method addresses the shortcomings of the 'dark channel prior' dehaze method by Kaiming He et al. Firstly, for the scenario of 'dark channel prior' failure, the 'linear stretching' method is proposed to estimate the dark channel after fog removal, then fuse the two 'dark channel' images, and the modified transmissivity is finally obtained. Secondly, because the introduction of the 'Guided Filter' increases the computation of the algorithm, this paper raises the 'pixel-based transmissivity estimation' proposed to remove the 'Guided filter', which not only saves hardware resources, but also eliminates the 'block effect' and 'Halo effect'.